ridm@nrct.go.th   ระบบคลังข้อมูลงานวิจัยไทย   รายการโปรดที่คุณเลือกไว้

Prediction of the extent of protein secondary structures using neural networks

หน่วยงาน จุฬาลงกรณ์มหาวิทยาลัย

รายละเอียด

ชื่อเรื่อง : Prediction of the extent of protein secondary structures using neural networks
นักวิจัย : Jurairat Phromjai
คำค้น : Proteins , Amino acids , Neural networks (Neurobiology) , Neural networks (Computer science)
หน่วยงาน : จุฬาลงกรณ์มหาวิทยาลัย
ผู้ร่วมงาน : Lerson Tanasugarn , Chulalongkorn University. Graduate School
ปีพิมพ์ : 2541
อ้างอิง : 9743324232 , http://cuir.car.chula.ac.th/handle/123456789/10174
ที่มา : -
ความเชี่ยวชาญ : -
ความสัมพันธ์ : -
ขอบเขตของเนื้อหา : -
บทคัดย่อ/คำอธิบาย :

Thesis (M.Sc.)--Chulalongkorn University, 1998

We present a method for predicting protein structures based on a digital computer of neural networks. The neural networks learned from existing protein how to predict the secondary structure of amino acid sequences. The amino acid properties of amino acids such as hydropathy, hydrophobicity, helical tendencies and amino acid side chain properties were used as input vector. These properties were coded into the amino acid sequences and used as input patterns for both training and testing. Seventy amino acid sequences and twenty-eight amino acid sequences from different proteins were used for training and testing respectively. The percent predictions accuracies of the existence of helix, sheet and turn structures using in the same network were lower than the prediction from separate networks. Each property gave the highest prediction accuracies for helix structure prediction. Properties can be ranked by their abilities to predict protein secondary structures as follow: (1.1) Amino acid side chain properties gave the highest accuracy for the prediction of the existence of helix, sheet turn in the same network, the existence of sheet and turn structure, percent helix, sheet (3 groups) and percent helix (2 groups). (1.2) Hydropathy (2 groups) gave the highest accuracy for the prediction of the existence of helix structure, percent helix and percent sheet (6 groups), percent turn (3 groups) and percent sheet (2 groups). (2) Hydropathy (7 groups) gave the highest accuracy for the prediction of percent sheet and turn (6 groups) and percent sheet (3 groups). (3) Hydrophobicity gave the highest accuracy for the prediction of percent helix (2 groups). The range of percent accuracy prediction from all properties for helix, sheet and turn were between 85-100%, 70-85% and 45-70% respectively. The range of percent accuracy from all properties for predictions of percent helix, sheet and turn (6 groups of 0%, 1-20%, 21-40%, 41-60%, 61-80% and 81-100%) were 35-65%, 30-50% and 25-50% respectively. Teh percent accuracies for percent helix, sheet and turn (3 groups of 0%, 1-50% and 51-100%) were 65-85%, 60-80% and 50-65% respectively. The percent accuracies for percent helix and sheet (2 groups of < 15% and > 15%) were 60-80% and 60-75% respectively. The percent of secondary structure prediction is useful for the folding classes prediction.

บรรณานุกรม :
Jurairat Phromjai . (2541). Prediction of the extent of protein secondary structures using neural networks.
    กรุงเทพมหานคร : จุฬาลงกรณ์มหาวิทยาลัย.
Jurairat Phromjai . 2541. "Prediction of the extent of protein secondary structures using neural networks".
    กรุงเทพมหานคร : จุฬาลงกรณ์มหาวิทยาลัย.
Jurairat Phromjai . "Prediction of the extent of protein secondary structures using neural networks."
    กรุงเทพมหานคร : จุฬาลงกรณ์มหาวิทยาลัย, 2541. Print.
Jurairat Phromjai . Prediction of the extent of protein secondary structures using neural networks. กรุงเทพมหานคร : จุฬาลงกรณ์มหาวิทยาลัย; 2541.